Forearm Orientation Invariant Analysis for Surface Myoelectric Pattern Recognition

In this work, we have performed a forearm orientation invariant analysis for artificial wrist movement classification on a specific subject using the surface electromyography (sEMG) signal. The muscle contraction effect has been included to incorporate the variation in force levels by the subject user. The Time-Domain Power Spectral Density (TD-PSD) predictors have been considered, which assists the real-time operation of the controller. Our main focus in this paper is to develop an optimum model, independent of the forearm orientation angle which will map the sEMG signals captured through the sensors on the forearm to the correct hand movement response. This is useful in developing intelligent artificial myoelectric controllers for both healthy patients as well as amputees requiring an intelligent artificial hand. It is found that the extreme forearm orientation angles (0° and 180°) have a significant improvement (> 15 %) in the testing accuracy compared to the other combinations for orientation invariant analysis.